Compaq redux: implicit taxes and the question of pre-tax profit.
Until recently, (1) more ink was probably spilled over the cross-border, dividend-stripping transactions in Compaq v. Commissioner (2) and IES Industries v. Commissioner (3) than over any other tax shelter litigation. (4) In both cases, the appellate courts reversed the trial courts' decisions in favor of the government, thereby granting the taxpayers the tax benefits they sought.
One reason why those cases have received and continue to receive so much attention is because the commentators are divided into two opposing camps. One group argues that the transactions were blatant, abusive tax shelters that the courts should have struck down. The other group argues that the transactions were economically profitable arbitrage transactions that deserved respect from the courts.
The debate is so vigorous because neither side has been able to come to grips with the other side's strongest argument. In spite of their best efforts to show otherwise, (5) the defenders of the circuit courts' opinions have not been able to demonstrate a convincing economic justification for the transactions other than tax reduction. (6) Similarly, critics of those opinions have not been able to rebut the claim that the transactions generate pre-tax profits. (7)
The debate, thus, strikes at the heart of anti-abuse jurisprudence, which for many years has relied heavily on the concept of pre-tax profit. (8) Over the years, courts have developed a series of doctrines that supplement statutory and administrative authorities. One of those doctrines is the economic substance doctrine.
In the conjunctive form in which it is usually stated, the economic substance doctrine requires both that the transaction has non-tax consequences and that the taxpayer has a non-tax motive. Although the pre-tax profit test plays a role in both prongs of the economic substance doctrine, it is frequently the centerpiece of the objective prong. A financial transaction that has a guaranteed pre-tax profit (appropriately measured) improves the taxpayer's economic situation and so it has a positive non-tax consequence. Conversely, a financial transaction that has a guaranteed pre-tax loss worsens the taxpayer's economic situation. If such a transaction produces a substantial tax benefit (and if the taxpayer entered into it with knowledge of that result), then the taxpayer must have only a tax motive. (9) Thus, in the context of financial arbitrage transactions, courts use the pre-tax profit test in order to separate economic arbitrages (the tax consequences of which are respected) from tax arbitrages (the tax consequences of which are not respected). (10)
The rationale for and the logic behind the pre-tax profit test is that any system as complicated and extensive as the U.S. tax law will contain numerous inconsistencies. When these inconsistencies are aggressively exploited, they allow taxpayers to reduce their taxes with very little consequence to the risk they bear, their cash flows, or their economic interests. The availability of such tax shelters, according to this line of reasoning, is outside the scope of congressional intent. Accordingly, courts will strike down transactions that fail the pre-tax profit test. (11)
With this history as background and precedent, the taxpayers and the commentators who sided with them argued that the transactions in Compaq and IES Industries should be respected and the taxpayers' claimed tax benefits allowed. They emphasized the point that the parties, at the time they entered into the transactions, had a reasonable expectation (and possibly a guarantee) of a before-tax profit from the transactions. The Fifth and Eighth Circuit Courts of Appeal agreed. Conversely, the government and the commentators who sided with them argued that the transactions were abusive tax shelters. In their view, the transactions--even though they yielded pre-tax profits--were nothing more than sales of U.S. foreign tax credits to domestic companies. (12) The trial courts agreed.
Compaq and IES Industries thus threaten to confuse tax shelter jurisprudence (even more deeply than it already is). The transactions appear to be tax shelters because they lack a nontax justification. At the same time, they pass the pre-tax profit test. Accordingly, several commentators argue that the pre-tax profit test needs to be abandoned or modified to reach such transactions. The problem and cause for concern is that it is not clear what test will replace the pretax profit test if that test is discarded. Some commentators favor a standard for abusive tax shelters along the lines of "we know it when we see it." (13) Many tax practitioners and commentators might be fine with using such a broad and indeterminate standard if the final decisions are to be made by the Tax Court. (14) Fewer would be prepared to grant such broad discretion to generalist courts, whether as trial courts or when reviewing the Tax Court. (15) Fortunately, Congress was able to avoid pushing the courts to discard the pre-tax profit test, at least for now, with a statutory fix that addressed the specific transaction at issue. (16) However, such a fix only eliminated the immediate problem raised by cross-border, dividend-stripping transactions. A similar problem might arise in the future with other cross-border, stripping transactions. (17) Furthermore, if Compaq and IES Industries reveal the existence of a fissure in anti-abuse jurisprudence, Congress's statutory fix does not close it, but merely masks it.
A deeper and more detailed resolution of the fundamental issue in Compaq and IES Industries is therefore warranted. Such an exercise is not only of historical significance: the issues raised by Compaq and IES industries are still before the federal courts. The federal government having lost Compaq in the Fifth Circuit and IES Industries in the Eighth Circuit is pursuing similar cases in other circuits. (18) Also, the cross-border, dividend-stripping transactions in Compaq and IES Industries are not the only cross-border, stripping transactions that give rise to foreign tax credits. How to handle such transactions is an important issue and is likely to remain so for some time. In addition, a broader resolution of the issues raised in Compaq and IES Industries might tell us something about our tax system that we have overlooked.
II. THE TRANSACTIONS
The transactions in Compaq and IES Industries are very similar. I will use the Compaq transaction to illustrate. The numbers are approximate and have been rounded for ease of discussion. On September 16, 1992, Compaq purchased and later that same day sold 10 million shares of Royal Dutch Shell. (19) The trades, which were made using American Depository Receipts (ADRs), were executed on the New York Stock Exchange (NYSE). (20) Compaq paid $887.5 million for the Shell ADRs and sold them for $868.4 million. Although Compaq held the shares for only about one hour, the transactions were timed so that Compaq would be the owner of record for Shell's October 1992 dividend. On October 2, 1992, Shell declared a dividend of $2.25 a share. Thus, Compaq was entitled to receive a $22.5 million dividend from Shell (the gross dividend).
Because Shell is a Dutch company, all dividends paid by Shell to its shareholders are Dutch source income. Under the terms of the tax treaty between the United States and the Netherlands, the Netherlands imposed a 15 percent withholding tax on dividends received by U.S. residents from Dutch companies. Thus, Compaq incurred $3.4 million in Dutch taxes on the $22.5 million dividend. (21) The Dutch tax was remitted by Shell before the dividend was paid to Compaq. The payment, however, was on behalf of Compaq and discharged its tax obligation. Accordingly, after remitting the Dutch withholding tax, Shell paid Compaq $19.1 million (the net dividend).
As a U.S. company, Compaq is taxable in the United States on its worldwide income from all sources. (22) Thus, Compaq reported the $22.5 million gross dividend from Shell as gross income on its 1992 U.S. tax return. (23) Because Compaq was subject to tax at the top U.S. corporate tax rate of 34 percent, (24) that income attracted $7.7 million of tax. (25) In addition, Compaq reported a short-term capital loss of $20.5 million from its sale of Shell stock. (26) Because Compaq had other capital gains that it offset with that loss, the Shell loss reduced Compaq's U.S. federal income taxes by $7 million. (27) Thus, the Shell transaction increased Compaq's U.S. tax obligation by $0.7 million the difference between $7.7 million and $7 million. (28)
Furthermore, the United States permits Compaq to receive a tax credit for withholding taxes paid to the Netherlands. (29) Accordingly, Compaq claimed a U.S. foreign tax credit for the $3.4 million in foreign withholding taxes paid to the Netherlands on its Shell dividend.
The above transaction, however, was not costless. Compaq incurred expenses of roughly $400,000 and paid Twenty-First Securities, the promoter, a fee of $1 million. (30)
Assembling all of the pieces of the transaction, Compaq's proceeds from the sale of the stock and the dividend total $890.2 million. Its cost of purchasing the stock was $888.9 million. The difference--$1.3 million--is Compaq's net after-tax profit from the transaction.
Table 1 presents Compaq's net cash flows from its transactions in Shell stock. The second column (labeled "Before Taxes") gives the cash inflows and outflows ignoring Compaq's tax payments. The third column (labeled "After Foreign Taxes") takes into account Compaq's tax payments to the Netherlands, but ignores its payments to the United States. The final column (labeled "After All Taxes") incorporates Compaq's tax payments to both the Netherlands and the United States (including the value of the foreign tax credit).
As Table 1 illustrates, Compaq earned an after-tax profit of $1.3 million from its brief transactions in Shell ADRs. That after-tax profit arose because when Shell went from trading cum dividend to ex dividend the price of the Shell ADRs dropped not by the gross amount of the dividend ($22.5 million), but by only the net amount of the dividend ($19.1 million). That suggests that the marginal holder of Shell's ADRs--the holder who sets the market price--could not use the foreign tax credit generated by the Dutch withholding tax. For such an investor, the dividend was worth only $19.1 million, which is why the stock dropped by that much when it went ex dividend.
In contrast with such holders, Compaq valued the dividend at its full $22.5 million. That is because Compaq had foreign source income on which it was paying U.S. tax, and so it could use the full value of the foreign tax credit to reduce its tax payments. (31) In addition, Compaq also had substantial capital gains (at least $20.5 million), which it could offset with the capital losses from its Shell transactions. Only because Compaq had substantial amounts of both foreign income on which it was paying U.S. federal income tax and capital gains not otherwise offset by capital losses did Compaq value the dividend at its gross amount, instead of its net amount. (32) Viewed from this perspective, Compaq paid $1.4 million in fees and expenses and $0.7 million in federal income taxes in order to receive a foreign tax credit worth $3.4 million. Thus, Compaq came out ahead by $1.3 million.
III. THE LITIGATION
The Internal Revenue Service (Service) attacked both Compaq's and IES Industries' tax treatments of their cross-border, dividend-stripping transactions on a variety of grounds. The Service disallowed the claimed foreign tax credits, capital losses, and deductions, and it sought to recover penalties from the taxpayers. The parties disagreed with the Service and went to court. Compaq went to the Tax Court, and IES Industries went to federal district court in Iowa. The district court in IES Industries held the transaction to be a sham with little analysis of either the transactions or the law. (33) The Tax Court in Compaq also disallowed the transactions, but on the grounds that they lacked economic substance. (34) In applying the pre-tax profit requirement of existing anti-abuse law, the Tax Court applied that test after deducting the Dutch withholding taxes paid by Compaq, but before deducting Compaq's U.S. taxes and adding back its foreign tax credits. (35) That is to say, Compaq's acquisition cost before taking into account U.S. taxes--$888.9 million--exceeded its revenue before taking into account U.S. taxes--$887.5 million. (36) Thus, as illustrated in Table 1, column 3, Compaq had a loss after Dutch taxes, but before U.S. taxes, of $1.4 million. Accordingly, the Tax Court disallowed the foreign tax credit because Compaq was certain to have a pre-tax loss from the transactions.
On appeal, however, the Eighth Circuit in IES Industries and then the Fifth Circuit in Compaq reversed the trial courts' decisions and held for the taxpayers. The main grounds on which the circuit courts based their decisions were the same in both cases. The Fifth and Eighth Circuits both rejected the Service's contention that pre-tax profit should be determined after the payment of the Dutch withholding tax, but before the payment of U.S. taxes and the receipt of foreign tax credits. Instead, both the Fifth and Eighth Circuits reasoned that foreign and domestic taxes should be treated alike. (37) Accordingly, in computing pre-tax profit, the courts ignored both sets of taxes. (38) Thus, looking at the cash receipts and disbursements from Compaq's transactions (and ignoring taxes), Compaq paid out $888.9 million (39) and received in exchange $890.9 million. (40) Thus, Compaq's net cash flow from the transactions before taxes was a profit of $2 million. In terms of Table 1, the circuit courts held that the lower courts erred in assessing pre-tax profit using the second column rather than the first. Looking at the first column of Table 1, the Fifth Circuit found for Compaq because the transactions produced a guaranteed before-tax profit for the taxpayer of $2 million. Thus, the Fifth Circuit permitted Compaq to use its foreign tax credits from the Shell transactions to offset its U.S. tax payments. (41) The Eighth Circuit held similarly. (42)
IV. THE LITERATURE
The circuit courts' opinions in Compaq and IES Industries unleashed a storm of commentary, much of it highly critical. Daniel Shaviro and David Weisbach describe the appellate courts' opinions as "appalling," "confused," "embarrassing," and "misguided" and the judges who authored them as lacking any "glimmer of comprehension" of both the pre-tax profit test and tax shelters. (43) David Hariton asserts that the test the appellate courts applied in Compaq and IES Industries were "all wrong." (44) Lee Sheppard declares that the Compaq opinion "is just plain wrong." (45) Mitchell Kane, although more restrained in his language, also believes Compaq and IES Industries were wrongly decided. (46) In the views of all of these authors, the underlying cross-border, dividend-stripping transactions were abusive tax shelters that the courts should have struck down.
William Klein and Kirk Stark are less certain. Although they express some reservations, they argue that Compaq's and IES Industries' cross-border, dividend-stripping transactions were legitimate business transactions--a form of economic arbitrage--and the tax benefits deserved to be respected. (47) A somewhat similar view is expressed by Alan Weiner. In a letter in Tax Notes in response to Hariton's article, Weiner argues that the purpose behind Compaq's transactions was to benefit from a pricing inefficiency in the market for Shell ADRs. (48) Indeed, my guess is that Weiner's view--that the transactions were business-driven, not tax-driven, because the transactions produced a guaranteed pre-tax profit--is the predominant one among tax practitioners.
Yet the claim that what is occurring is economic, not tax arbitrage, is not convincing. It is difficult to believe that the stock market misprices such a simple and common transaction as a stock paying a known cash dividend by 15 percent of the gross dividend. Moreover, the stock price drop when the Shell ADRs went ex dividend ($19.1 million) was precisely the net dividend. Such a price drop is precisely what would be expected if the market were efficient and the marginal holder could not get any benefit from the Dutch withholding tax. (49)
The commentators critical of the circuit courts' opinions in Compaq and IES Industries also faced a challenge--the cross-border, dividend-stripping transactions in those cases were profitable before taxes (see Table 1, column 2) and pre-tax profit has long been the cornerstone of anti-abuse jurisprudence. Many of these commentators responded by recommending that the pre-tax profit test either be modified or abandoned and replaced with a different test that would do a better job of reaching cross-border, dividend-stripping and other similar tax arbitrages. (50)
V. THE RESOLUTION
The confusion engendered by Compaq and IES Industries can be traced back to a mistaken conclusion with which nearly everyone agrees--that the cross-border, dividend-stripping transactions in those cases generated pre-tax profits. The key to understanding the pre-tax profit issue in Compaq and IES Industries is drawn from an important concept in public finance--the notion of implicit taxes. Because the courts, the parties, and commentators all missed the role played by implicit taxes in the Compaq and IES Industries transactions, they also applied the pre-tax profit test incorrectly. Once implicit taxes are taken into account, it is clear that the transaction produced a pre-tax loss, not a pre-tax profit. Thus, had the appellate courts correctly applied the pre-tax profit test, by assessing profit before all taxes, they would have held for the government.
A. Implicit Taxes and Clientele Effects
Start with a simple example. Consider a taxpayer who holds a (riskless) zero-coupon corporate bond that she purchases for $1000 and that pays $1100 at maturity in exactly one year. (51) The interest on such a bond is included by the taxpayer in her income. Thus, if a taxpayer purchases for $1000 a corporate bond that pays $1100 at maturity, the taxpayer includes $100 interest in her income. If she is taxed at 40 percent, she pays $40 in tax and is left with $1060 after tax. Such a taxpayer earns 10 percent interest before tax and 6 percent interest after tax on her investment. Alternatively, if the taxpayer purchases a municipal bond, she does not have to include the interest she receives in income. If such a bond has the same terms (and is also risk free), the taxpayer receives $1100, pays no tax and so ends up with $1100, which is $40 more than she has with the corporate bond. Expressed differently, she receives 10 percent both before tax and after tax with the municipal bond compared to 10 percent before tax and 6 percent after tax with the corporate bond. Thus, if both investments are available to the taxpayer, the taxpayer will prefer the municipal bond to the corporate bond.
Of course, other taxpayers with marginal tax rates above zero will also prefer the municipal bond to the corporate bond. As long as the terms were the same, no taxpayers (with positive tax rates) will want to hold corporate bonds. All will prefer to hold municipal bonds. Thus, we can expect competition among investors and issuers to drive down the interest rate on municipal bonds and drive up the interest rate on corporate bonds. To keep things simple, assume that there is an elastic supply of corporate bonds that pay 10 percent interest (so a decline in demand will not increase their interest rate) and a small and not as elastic supply of municipal bonds. Taxpayers will therefore bid down the interest rate on municipal bonds to 6 percent. At this point, taxpayers in the 40 percent bracket will be indifferent between the two bonds. (52)
Economists use the phrase "implicit tax" to refer to the reduction in the return from holding a tax favored investment. There are three alternative ways to quantify the implicit tax on the municipal bond. First, the implicit tax can be described as the increase in the issue price of the municipal bond, assuming that the municipal bond and the otherwise identical corporate bond pay the same $1100 at maturity. Such a municipal bond will cost $1037.74. (53) Viewed in this manner, the implicit tax is $37.74. Second, the implicit tax can be described as the decrease in the payment on the municipal bond at maturity assuming that the municipal bond and the otherwise identical corporate bond cost the same $1000 at issuance. Such a municipal bond will pay $1060 at maturity. (54) Viewed in this manner, the implicit tax is $40. (55) Third, the implicit tax can be described as a reduction in the interest rate paid by the municipal bond. Viewed in this way, the implicit tax is a 4 percent reduction in the interest rate, from 10 percent down to 6 percent. (56) All three methods are alternative, but equivalent, methods of describing the same phenomenon.
Closely related to the concept of implicit tax is the notion of an implicit tax rate. The implicit tax rate is the amount of implicit tax expressed as a percentage of the total before-tax return. That is the most common method of talking about implicit taxes because that method can be readily used to assist investors in directing their investments among alternative investments. In this example, a taxpayer in the 40 percent bracket pays explicit tax of 40 percent on her interest from holding the corporate bond. Taxpayers taxable at different rates will pay explicit taxes at different amounts and rates. Thus, a taxpayer taxable at 30 percent will pay explicit tax at 30 percent and a taxpayer taxable at 50 percent will pay 50 percent. However, all taxpayers who hold the municipal bond pay implicit tax at the same rate--40 percent. That is because all holders receive a 6 percent after-tax return on municipal bonds instead of a 10 percent before-tax return on corporate bonds. Because every investor could purchase corporate bonds paying 10 percent, every holder of municipal bonds pays implicit tax at a rate of 40 percent. That forty percent implicit tax rate is calculated as follows:
40 percent = (10 percent - 6 percent)/10 percent,
where the numerator is the difference in return between the tax-free municipal bond and the equivalent taxable corporate bond and the denominator is the return on the taxable corporate bond. The denominator is the return on the corporate bond, 10 percent, because the corporate bond represents the benchmark asset--the asset that is taxed in accordance with the dictates of the tax system and that is available to receive excess investment funds. (57)
In this simple example, each taxpayer decides whether to hold municipal bonds or corporate bonds by trading off implicit and explicit taxes. A taxpayer should choose the bond for which her total tax rate is the lowest. Thus, a taxpayer in the 20 percent tax bracket prefers to hold the corporate bond rather than the municipal bond. She is better off paying 20 percent explicit tax rather than 40 percent implicit tax. The situation is reversed for a taxpayer in the fifty percent bracket. She is better off paying implicit tax at 40 percent rather than explicit tax at 50 percent.
The basic principles carry over, but the arithmetic is messier, when the tax favored investment is not exempt from tax, but is instead taxed at a reduced rate. For example, assume that there is another class of bonds where the taxpayer is required to include only half of her income from the bond. From the perspective of a 40 percent taxpayer, she is taxed at 20 percent on the coupon from such a bond. If 40 percent taxpayers competed to acquire such bonds, they will bid up their price to $1023.26. (58) Alternatively, if such bonds each cost $1000, they will pay $1075 at maturity. Thus, such lightly taxed bonds generate a before-tax return of 7.5 percent.
As above, we can describe the implicit tax on the lightly taxed asset in three different, but equivalent, ways. The implicit tax is an increase in the purchase price of $23.26, a reduction in the payment at maturity of $25, or a reduction in the interest rate of 2.5 percent--from 10 percent to 7.5 percent. Expressed in terms of a tax rate, the implicit tax rate on the bond is 25 percent. (59)
Consider now what happens if there is still another class of bonds that is taxed more heavily than the benchmark asset. Assume, for example, that the holders of such bonds are taxed on double their income. Thus, if such a bond costs $1000 and pays $1100 at maturity, the taxpayer includes $200 in income. For a taxpayer in the 40 percent bracket, she pays $80 in tax and is left with $1020. That is $40 less than what she has with the same investment in a corporate bond. The 40 percent taxpayer will avoid such a bond, as will all taxpayers with positive tax rates, and so its price will drop. Assuming that investors in the 40 percent tax bracket set the price for such bonds, then the price of the heavily taxed bond will drop to $846.15 in order to provide the same after-tax return as a corporate bond. (60) At this price, the heavily taxed bond pays a stated rate of return of 30 percent. (61) We can call the increased return or reduced price on the bond an implicit subsidy or a negative implicit tax. It arises from the extra heavy explicit tax on the bond.
Once again, we can describe the implicit tax on the highly taxed asset in three different, but equivalent, ways. The negative implicit tax is a decrease in the purchase price of $153.85, an increase in the payment at maturity of $200, or a 200 percent increase in the interest rate--from 10 percent to 30 percent. Expressed in terms of a tax rate, the implicit tax rate on the bond is negative 200 percent. (62)
Returning to Compaq and IES Industries, the connection between the above example and those cases is made by asking the question, what is the cost of each of the differentially taxed bonds before taxes? Assume that the bonds all pay $1100 at maturity. The obvious answer is the issue price of each bond--$1037.74 for the untaxed bond, $1023.26 for the lightly taxed bond, and $846.15 for the heavily taxed bond. That answer, however, is wrong. Those prices are before explicit taxes, but they are after implicit taxes. As given in Table 2, the cost of each bond before all taxes, both explicit and implicit (and including negative implicit taxes), is $1000. In each case, the difference between the issue price and $1000 is an implicit tax. That implicit tax is positive for the untaxed ($37.74) and lightly taxed ($23.26) bonds and negative (-$153.15) for the heavily taxed bond.
B. Application to Compaq and IES Industries
Essentially, the same question is at the heart of Compaq and IES Industries. And it is the same sort of oversight, equating the cash purchase price with the before-tax cost, that is the source of the appellate courts' mistaken conclusions that the transactions in Compaq and IES Industries were profitable before taxes.
The appellate court in Compaq sought to determine whether Compaq had a before-tax profit on its transactions in Shell stock. That, in turn, depends on the pre-tax cost to Compaq of its Shell stock. Ignoring implicit taxes, the court took Compaq's out-of-pocket purchase price--$887.5 million--and added fees and expenses--$1.4 million--to arrive at Compaq's pre-tax cost of acquiring the shares--$888.9 million. Compaq's pre-tax revenue from the transactions is the sum of Compaq's sale price for its Shell shares ($868.4 million) and gross dividend ($22.5 million), which totals $890.9 million. Thus, because Compaq's pre-tax revenue from the transactions ($890.9 million) exceeded its pre-tax cost ($888.9 million) by $2 million, the transactions produced a pre-tax profit. The Fifth Circuit, therefore, held that Compaq's Shell transactions had economic substance and so it reversed the trial court, thereby allowing Compaq to take the tax benefits that it sought.
One of the points emphasized by Myron Scholes, Mark Wolfson, and their co-authors in their tax planning textbook, Taxes and Business Strategy, is that implicit taxes are real taxes, just as are explicit taxes. (63) Although the significance of implicit taxes is evident when a taxpayer is choosing between taxable and tax-exempt bonds, (64) implicit taxes are often overlooked by courts and commentators when applying legal principles. (65) Taking implicit taxes into account and recognizing that there can be (positive) implicit taxes as well as negative implicit taxes (implicit subsidies) leads to the conclusion that the transactions in Compaq and IES Industries do not generate pretax profits, but instead produce pre-tax losses.
Viewed from this perspective, Compaq did not pay $887.5 million before tax ($888.9 million after expenses) for its shares of Shell. Such a payment is not before all taxes because it is after the negative implicit tax (implicit subsidy) paid by Compaq that is a direct result of the withholding tax. The withholding tax suppressed the cum dividend price of Compaq's Shell stock by $3.4 million--the amount that Shell subsequently withheld and paid over to the Dutch taxing authorities on Compaq's behalf. Thus, the amount paid by Compaq before all taxes (including implicit and explicit taxes, whether positive or negative) for its Shell stock was $890.9 million. (66) Thus, Compaq's before-tax cost, including expenses ($1.4 million), was $892.3 million. That exceeded its before-tax proceeds of $890.9 million by $1.5 million. Thus, when proper account is taken of implicit taxes, the transactions produced a pre-tax loss, not a pre-tax gain. (67) See Table 3 below:
TABLE 3. COMPAQ'S BEFORE-TAX CASH FLOWS FROM ITS TRANSACTION IN SHELL STOCK. Before All Taxes After All Taxes (Both Explicit (Both Explicit Proceeds from Sale and Implicit) and Implicit) and Dividend: (in millions) (in millions) Sale $868.4 $868.4 Dividend 22.5 22.5 FTC -- 3.4 W/h Tax -- (3.4) US Tax -- (0.7) Subtotal $890.9 $890.2 Cost of Purchase: Purchase $890.9 $890.9 Fees 1.0 1.0 Expenses 0.4 0.4 Implicit Subsidy -- (3.4) Subtotal $892.3 $888.9 Net Profit ($1.4) $1.3
C. The Nature of the Arbitrage in Compaq and IES Industries
In order to see clearly the tax arbitrage in Compaq and IES Industries, recall that Compaq's $22.5 million Shell dividend was subject to $3.4 million Dutch withholding tax. That tax, which Shell remitted on behalf of Compaq, was the only tax Compaq owed to the Netherlands as a result of its transactions in Shell stock. Specifically, in calculating its Dutch taxes, Compaq could not deduct its $20.5 million capital loss on its Shell share. (68)
Although the Dutch withholding tax looks like an income tax, it is not a tax on income as economists ordinarily understand the term. Instead, it is a tax on a specific payment, a dividend, regardless of the taxpayer's income. The tax applies to all foreign holders, whether for-profit or non-profit, and without regard to the taxpayer's income. (69) That is to say, Compaq did not receive any Dutch tax benefits from its capital loss on its Shell shares. Furthermore, Compaq's liability for the tax was independent of the length of time it held the stock (as Compaq was counting upon). Instead, it was a direct result of owning the stock at a point in time--the moment of record for the payment of the dividend. (70) Thus, when the Shell stock went from trading cum dividend to ex dividend, it also went from trading cum withholding tax liability to trading ex withholding tax liability.
For holders of Shell stock who could use the full foreign tax credit and had no problem with the embedded capital loss (because they had capital gains that the losses could offset), the Dutch withholding tax was not detrimental. For such taxpayers, the Dutch withholding tax was offset by a dollar-for-dollar reduction in the tax paid to the U.S. government. Accordingly, such taxpayers valued the dividend at its gross amount. Conversely, for taxpayers who could not make any use of the foreign tax credit, the Dutch withholding tax was a detriment to the full extent of the tax. Accordingly, such taxpayers valued the dividend at its net amount.
If the market for the Shell ADRs contained many investors who could use the full foreign tax credit, then the market would value the dividend at its gross amount and the price of the stock would have dropped when the stock went ex dividend by the gross dividend. Alternatively, if the market contained few investors who could use the foreign tax credit in whole or part, then the market would have valued the dividend at its net amount and the price of the stock would have dropped when it went ex dividend by the amount of the net dividend. Moreover, such a tax would have reduced the sale price of the stock before the stock went ex dividend by the present value of withholding tax. (71)
Of course, the actual market for Shell ADRs in 1992 was composed of both types of investors--those that received full value for the foreign tax credit and those that received no value--and probably a few that received only some (expected) value for the foreign tax credit. It might, then, be thought that the market represented an average or midpoint of those investors. Equilibrium, however, in financial markets is determined not by the average, but by the marginal, investor. Because most investors likely would have received either no value or full value on the foreign tax credit, one or the other group of investors likely determined the market price for Shell ADRs when Compaq entered the market in 1992.
Moreover, evidence from the market suggests that the investors who paid the Dutch withholding tax, but did not receive any benefit from that tax set the price of Shell ADRs. That is to say, at the margin, prices were set by investors who could not use the withholding tax to offset other taxes. (72) When the Shell ADRs went from trading cum dividend to ex dividend, their price dropped not by the gross dividend, but by the net dividend. The price dropped by only the net dividend because the withholding tax had already depressed the cum dividend price of the ADRs by the (expected present value) of the withholding tax. (73)
It, therefore, follows that the arbitrages in Compaq and IES Industries are examples of clientele-based arbitrages. (74) The key to clientele-based arbitrage is the trade-off between implicit and explicit taxes. (75) Compaq and IES Industries were implicitly paid for taking on the obligation to explicitly pay the withholding tax on the dividend. That obligation would have otherwise fallen on the original owners of the stock. Those owners implicitly paid Compaq, IES Industries, and other market purchasers for paying that tax by selling them their stock at a discount to what the stock would have otherwise sold for without the Dutch tax. For Compaq and IES Industries, the reduced purchase prices they paid were negative implicit taxes, which in the case of Compaq was in the amount of $3.4 million. (76) For Compaq, which was excess limitation and could use the full foreign tax credit, the explicit tax on the transaction was substantially reduced by the credit--it was only $0.7 million. Thus, Compaq traded the $3.4 million negative implicit tax against a $0.7 million tax payment. After $1.4 million in fees, Compaq came out ahead by $1.3 million from its clientele-based arbitrage.
D. The Advantages from Measuring Pre-Tax Profit Before Implicit Taxes
In this section, I argue that assessing pre-tax profit before implicit taxes is consistent with the purpose behind the economic substance doctrine. I also argue that such a test has numerous advantages over the traditional method of assessing pre-tax profit, which ignores implicit taxes.
The logic behind the economic substance doctrine is that it separates transactions imbued with tax-independent considerations from those shaped solely by tax avoidance. (77) Accordingly, a transaction that is guaranteed to generate a profit is a business transaction, whereas one that is certain to produce a loss is a tax avoidance transaction. (78) Thus, the pre-tax profit test separates economic arbitrage from tax arbitrage transactions. Treating implicit taxes as taxes helps the pre-tax profit test to reach correct conclusions, whereas the failure to treat implicit taxes as taxes can cause that test to go astray. Moreover, the failure to take proper account of implicit taxes invites taxpayers to undertake tax arbitrages that will pass muster and to avoid those transactions--tax arbitrages and others--that will not. That incentive to find the viable tax arbitrages is the meta-lesson of Compaq and IES Industries. In both cases, the courts reached the wrong results because they ignored implicit taxes. Both courts mistakenly concluded that the transactions were economic arbitrages because they appeared to produce pre-tax profits when they were actually tax arbitrages because they produced pre-tax losses. And by doing so, they invited others to take advantage of that oversight.
Furthermore, the apparent failure to treat implicit taxes as taxes in applying the pre-tax profit test will produce wrong conclusions not only for cross-border, dividend-stripping transactions. It will produce wrong results in a broad range of transactions. Consider a very simple arbitrage transaction--borrowing and lending, where one side of the transaction is tax exempt and the other side is taxable. Assume tax-exempt debt pays 6 percent and taxable debt pays 10 percent, and both types of debt are riskless. A tax-exempt borrower can earn the 4 percent spread by borrowing tax free and purchasing taxable debt. As the pre-tax profit test is currently formulated and applied, that transaction passes muster because it generates a pre-tax profit equal to the 4 percent spread. Of course, the transaction is not a business transaction, but a tax arbitrage. (79) On a before-tax basis, it breaks even, but after-tax it yields 4 percent. Conversely, the opposite side of the same transaction carried out by a taxpayer in the 50 percent bracket would fail the pre-tax profit test. Such a taxpayer has a pretax loss equal to the 4 percent spread, but an after-tax gain of 1 percent. That is not right either. The taxpayers are really breaking even before tax. They are trading off implicit and explicit taxes in whatever way advantages them. A pre-tax profit test that accounts for implicit taxes correctly identifies both sides of the arbitrage; a test that ignores implicit taxes does not.
More generally, a pre-tax profit test that takes account of implicit taxes will treat both sides of clientele-based tax arbitrages as tax arbitrages. Unlike the existing pre-tax profit test, which treats one side of such arbitrages as tax arbitrage and the other side as economic arbitrage, my proposed test will treat both sides the same. That is because a pre-tax profit test that treats implicit taxes as taxes tends toward treating tax arbitrages as zero-profit transactions before taking account of fees and expenses. Thus, transactions with substantial fees and expenses will generate pre-tax losses unless they also produce economic benefits larger than the fees and expenses. (80)
Furthermore, a pre-tax profit test that takes account of implicit taxes will yield the same pre-tax profit when the market values the dividend at its gross amount, so there is no implicit tax. Some U.S. investors, such as Compaq and IES Industries, were able to use the Dutch withholding tax to offset a portion of their federal income tax liability, and thus they were able to avoid the impact of the tax. Because the United States has a worldwide tax system, the United States provides U.S. taxpayers with a foreign tax credit to alleviate what would otherwise be double taxation. Accordingly, taxpayers with less than the maximum amount of foreign tax credits allowed by law (81) and capital gains can use the foreign tax credit to offset U.S. taxes and the capital losses to offset capital gains. (82) Such taxpayers will be made no worse off by the Dutch withholding tax assuming that they paid as much for their stock as they received when they sold it and from the (gross) dividend. If such taxpayers determined the market price of Shell stock cum dividend in 1992, the dividend would have been worth its gross amount (not its net amount). In that case, the price of Compaq's Shell stock immediately before going ex dividend would have been $890.9 million and there would be no (negative) implicit tax. Under these circumstances, the transaction would still have produced a $1.4 million loss before tax. (83) A before-tax profit test that treats implicit taxes as taxes views the pre-tax profit on this transaction and on the actual transaction in Compaq as equal--both produce a before-tax loss equal to fees and expenses. In contrast, the traditional way of applying the pre-tax profit test views these transactions differently. The former is viewed as economic arbitrage because there is a pre-tax profit; the latter is viewed as (unprofitable) tax arbitrage because there is a pre-tax loss. However, all that has changed is that the marginal purchaser can now use the foreign tax credits generated by the Dutch withholding tax.
E. How the Role of Implicit Taxes in Pre-Tax Profits Was Missed
If the proper way to think about pre-tax profit is before both implicit and explicit taxes, then how did the parties, the courts, and the commentators all miss that argument? I believe there are two principal reasons for that oversight. First, although public finance economists have long known about implicit taxes, the tax law has yet to incorporate implicit taxes into tax doctrine. Public finance concepts tend to make their way into the tax law slowly. (84) And in many instances, those concepts are incorporated into the law only in response to specific and widespread tax planning strategies that directly target a specific failure and are widely viewed as abusive. (85) Although implicit taxes have not yet raised such a high level of concern, they might be starting to as evidenced by the attention the government is paying towards cross-border, stripping transactions. (86)
Second, the implicit tax arising in a cross-border, dividend-stripping transaction is largely concealed from view. By their very nature, implicit taxes--as their name suggests--are hidden taxes. However, relative to other implicit taxes, the implicit taxes in Compaq and IES Industries were especially well concealed. The usual way of thinking about implicit taxes is as increases in the prices of assets that are taxed less heavily than other assets. It is far less common to speak of negative implicit taxes, which are reductions in the prices of assets that are taxed more heavily than other assets. (87) Moreover, the excess taxation that occurs with cross-border, dividend strips is further obscured because it is the asymmetric treatment of the dividend (taxed abroad) and the corresponding capital loss (not taxed abroad) that is the source.
VI. THE GOVERNMENT'S ARGUMENTS IN COMPAQ AND IES INDUSTRIES AND COMMENTATORS' PROPOSALS TO MODIFY OR ABANDON THE PRE-TAX PROFIT TEST
Commentators and the government in Compaq and IES Industries have offered several proposals to modify or replace the pretax profit test. All of these proposals would have allowed the courts to reach different results in Compaq and IES Industries. In this Part, I argue that incorporating implicit taxes into the pre-tax profit test is preferable to either modifying or eliminating the pre-tax profit test entirely. I conclude this Part with a brief examination of the courts' discussion of risk reduction and commentators' reactions to that discussion.
A. The Argument that Pre-Tax Profit Should Be Calculated After Foreign Taxes
It is important to distinguish my argument--that the cross-border, dividend-stripping transactions did not produce a pre-tax profit--from the argument made by the government to that same effect in Compaq and IES. Facially, my argument and the government's argument are similar--both reduce the sum of the sales price and dividends relative to the purchase price by the amount of Dutch withholding tax ($3.4 million)--but the similarity ends there. The two sets of arguments are based on conceptually different foundations and in other situations will yield conflicting results.
I argued above that Compaq's before-tax cost of purchasing its Shell ADRs should be increased by the amount of the Dutch withholding tax. That is because the withholding tax depressed Compaq's purchase price by the amount of the tax. Such a reduction in the purchase price was an implicit (negative) tax. That tax should have been excluded from a before-tax analysis of Compaq's profit, thereby increasing its before-tax purchase price by $3.4 million, and resulting in a before-tax loss.
In contrast, the government argued in Compaq that the proceeds from the dividend should be reduced by the amount of the withholding tax. The government gave two reasons for that position. First, the government argued that the dividend ($22.5 million) should be reduced by the Dutch withholding tax ($3.4 million) because the tax was remitted by Shell before it paid the remainder of the dividend ($19.1 million) to Compaq. Second, the government argued that foreign taxes ($3.4 million) should be treated as an expense in assessing whether a transaction generates a pre-tax profit.
The appellate courts had little trouble rejecting the government's first argument, which would have made the tax result turn on the method of collection as opposed to the economic consequences of the transactions. Under that argument, if Compaq had remitted the tax itself, after receiving a dividend payment from Shell, the legal result would be different. Because the before-tax dividend would then be $22.5 million, not $19.1 million, the transaction would produce a pretax profit. Set within a body of tax doctrine that purports to tax transactions on their substance, not their form, the government's proposed rule would be out of place. Also, without more, the suggestion that courts should focus on who collects the withholding tax and transmits it to the authorities comes across as unprincipled, ad hoc, and driven exclusively by the intended result.
The second reason offered by the government for why Compaq's pre-tax profit should be calculated after payment of the Dutch withholding tax--that foreign taxes should be treated differently than domestic taxes--received little judicial attention. The appellate courts spent little time disposing of this argument since the government offered no precedent and presumably few, if any, reasons in support.
Although the government did not pursue it, Shaviro and Weisbach vigorously took up the government's argument that pre-tax profit should be calculated after foreign taxes (but before U.S. taxes). They argue that countries need to protect their own tax systems from abusive tax shelters and treating foreign taxes as before-tax expenses is consistent with that position. (88) They appear to have had an impact: in 2006, the federal government expressed its intention to challenge cross-border, dividend-stripping transactions outside the Fifth and Eighth Circuits using the argument that pre-tax profit should be calculated after foreign taxes. (89) Such an approach, however, is problematic.
The idea that we should treat all foreign taxes as an expense in applying the pre-tax profit test has a beggar-thy-neighbor quality and conflicts with broad principles of tax neutrality. (90) It can also produce undesirable and unintended results. For example, if a cross-border tax shelter generated large tax savings abroad and in the United States, those foreign tax savings would increase the "before-tax" profit from the transaction, and so could turn a transaction that would otherwise be classified as a tax shelter into a legitimate transaction. (91) Treating foreign taxes as occurring before-tax can also turn business transactions with no evident tax motivation or benefit into tax shelters. For example, as Shaviro and Weisbach themselves point out, borrowing money at 7 percent that generates a U.S. tax deduction in order to invest that money abroad at 10 percent in a country that imposes tax at 35 percent would qualify as a tax shelter. (92) Another failing of such an approach is that it would treat identical domestic and cross-border transactions with otherwise identical tax consequences differently. (93)
Although the argument that foreign taxes should be treated as expenses for the purpose of applying the pre-tax profit test is conceptually problematic and has twice been rejected by the circuit courts, the government continues to press it. In 2006, a Chief Counsel Advice asserted the government's litigation position in those cases. The government's argument is as follows:
In reaching this conclusion [that the taxpayer did not anticipate a pre-tax profit], we note that Taxpayer resides in the Second Circuit, which is not bound to follow the portions of the Fifth and Eight Circuit decisions in Compaq and IES that conclude that, for purposes of the economic substance analysis, the pretax profit should be determined without taking into account the cost of the expected foreign tax on the transaction. The failure of those courts to subtract the economic cost to the taxpayer of the foreign tax undermines the test's purpose of determining whether the taxpayer had a real potential for profit apart from the transaction's U.S. tax benefits (i.e., the foreign tax credit). In determining whether a taxpayer had a reasonable possibility of economic profit, all of a taxpayer's items of income and expense must be taken into account. Economically, a foreign tax is no different from any other expense, and therefore foreign taxes are properly treated as a cost that reduces economic profit. If a taxpayer's return is negative before U.S. tax consequences are taken into account, as is the case here, it necessarily follows that the benefit of the foreign tax credit motivated the transaction. Ignoring foreign taxes as a cost in determining the pre-U.S. tax return on a transaction would deprive the economic substance test of the very measure it is designed to illuminate--the return on the transaction prior to the claimed U.S. tax benefits. (94)
Although carefully drafted and argued, the government's statement represents a shift in the purpose behind the pre-tax profit test (and the economic substance doctrine). Traditionally, the purpose behind the test (and doctrine) has been seen as separating purely tax-motivated transactions (the tax benefits of which are not respected) from business transactions (the tax benefits of which are respected). Most investors treat foreign and domestic taxes the same. They care about how much tax they pay, not to whom they pay it. Thus, foreign and domestic taxes are treated on par. Conversely, treating foreign and domestic taxes differently undercuts the ability of the pre-tax profit test to separate business-motivated activity from tax-motivated activity.
In contrast with the government's position, the argument that courts should calculate before-tax profit before implicit taxes as well as before explicit taxes does not depend upon formalities. The result of the test does not depend upon who remits the tax to the Dutch treasury or even whether the tax is imposed by foreign or domestic law. Either way, the result is the same. (95) Moreover, and most important, incorporating implicit taxes into the pre-tax profit test does not undercut the purpose behind the pre-tax profit test (and the economic substance doctrine). In contrast, it advances the test's purpose of separating business transactions from tax-driven transactions.
B. Commentators' Arguments to Modify or Replace the Pre-Tax Profit Test
The commentators take a different tact than the one taken by the government. Instead of trying to shoehorn the transaction into the existing pre-tax profit test by arguing that foreign taxes should be treated as expenses, they recommend instead that the pre-tax profit test be either modified or abandoned.
For example, Kane proposes:
[I]n transactions involving the acquisition of assets that will produce income subject to foreign withholding tax, the economic substance test should be applied not merely by comparing the pre-tax and after-tax profit on the taxpayer but also by determining whether there is any after-tax profit from the transaction in excess of the hit taken by the Treasury." (96)
As applied to Compaq, the cost to the U.S. Treasury was $2.7 million--the difference between Compaq's $3.4 million foreign tax credit and the additional $0.7 million tax paid by Compaq on the additional $2 million of income reported by Compaq. (97) Because that amount exactly equals the sum of Compaq's purported pre-tax profit of $1.3 million and its fees and expenses of $1.4 million, the transaction would fail Kane's revised pre-tax profit test. (98)
In effect, Kane's revised pre-tax profit test reframes the question whether there is a pre-tax profit by looking at the taxpayer and the government together instead of solely at the taxpayer. If the taxpayer's asserted pre-tax gain is less than or equal to the government's tax cost, then there is not really a pre-tax profit from the transaction, but actually a pre-tax loss. In cross-border, dividend-stripping cases, such as Compaq and IES Industries, that test, I believe, will give the same result as explicit incorporation of implicit taxes into the pre-tax profit test.
There are, however, several problems with such a revised pre-tax profit test. First, it comes across as ad hoc. Kane offers no economic foundation for such a test. (99) Second, there are no clear limits where the test applies. Kane proposes that the test be applied to transactions with withholding taxes, but there is no evident reason for such a limit. Such a limited test, however, is problematic because it threatens to treat cross-border transactions differently than their domestic twins. (100) Yet there must be limits to the reach of Kane's proposed pre-tax profit test. The test, for example, does not work or help when the transaction in question is the trade-off from arbitraging the difference in tax rates between taxable and tax-exempt bonds, whether they produce foreign withholding taxes or not. Third, such a revised test only works when the counterfactual is readily available. In many cases, although not Compaq and IES Industries, there would likely be a lively debate over the proper counterfactual.
Another commentator, Hariton, goes even further. He argues that courts should abandon the pre-tax profit test altogether. Specifically, Hariton argues that the before-tax profit test will lead courts to uphold tax shelters that should be struck down when the taxpayer stuffs the transaction with income-producing assets. (101) He further argues that Compaq and IES Industries demonstrate that the before-tax profit test is the wrong test for just this reason.
Of course, Compaq and IES Industries do not reveal the failure of a before-tax profit test to distinguish legitimate business transactions from tax shelters. As I argue above, a proper application of the before-tax profit test that takes account of both explicit and implicit taxes, reveals the transactions as losers before all taxes. In addition, courts can conceptually separate out investments that are added to a tax shelter from the shelter itself and apply the before-tax profit test to the shelter itself. (102) Furthermore, there is a lacuna in Hariton's argument that Compaq was wrongly decided because the pre-tax profit test can be evaded by stuffing a transaction with income producing assets. The gap in Hariton's argument is that is not what happened in Compaq and IES Industries. In neither case did the taxpayer make a substantial investment for an appreciable length of time. The transactions took only about an hour and the funds were borrowed.
In addition to the possibility of stuffing transactions with income-producing assets, Hariton (in his earlier work) describes a second potential problem with the pre-tax profit test--the possibility that not all of the risk in a transaction will be eliminated. (103) In such a case, the taxpayer need not have a guaranteed profit or loss; instead, the taxpayer might have the possibility of achieving either a loss or a profit on a transaction. Of course, that was not the case with Compaq and IES Industries because the taxpayers eliminated all of their risk. (104) Nonetheless, because of these two possibilities, Hariton suggests a different test. He urges courts to replace the before-tax profit test with a test of whether the taxpayer has altered her economic position. (105) Under Hariton's proposed test, if the taxpayer has sufficiently altered her economic position, then she is entitled to the transaction's tax benefits.
The approach advocated by Hariton has won some adherents. For example, although Shaviro and Weisbach argue at length that the courts in Compaq and IES Industries should have accepted the government's argument that foreign taxes are an expense for the purpose of the pre-tax profit test (and thereby concluded that the transactions were unprofitable and hence the tax benefits should not have been respected), that is not their principal claim. Their principal claim is that any mechanical test for tax shelters is likely to fail to separate legitimate business transactions from tax shelters. (106) Thus, like Hariton, they advocate a flexible "know it when you see it" approach to tax shelters. (107) Similarly, Terrence Chorvat offers a theoretical argument why any mechanical test for tax shelters will not work, and so he also advocates a flexible approach. (108)
Whatever the merits to such a flexible test generally, an important end point is the pre-tax profit test. If a taxpayer enters into a transaction without incurring a net investment or any risk, and if that investment produces a positive guaranteed return, then the transaction is an example of economic arbitrage. In Hariton's language, the transaction changes the taxpayer's position (in a beneficial way) by increasing the taxpayer's wealth. Consequently, such a transaction is likely to be driven (at least in part) by business or economic considerations and so is deserving of respect. Alternatively, if the same transaction produces a guaranteed loss, then the transaction is not being driven by business or economic considerations (because it does not change the taxpayer's position in any beneficial way other than taxes) and so it is not worthy of respect. (109)
C. The Compaq Court's Discussion of Risk
In addition to arguing that the transaction did not generate a pretax profit, the government in Compaq also argued that Compaq's elimination of the risk from holding Shell ADRs was an independent and sufficient basis for the court to strike down the transactions. The Fifth Circuit rejected that argument. Instead, it concluded that it was a smart business practice for Compaq to eliminate its risk by locking in the resale price of their Shell ADRs. (110)
Several commentators were very critical of the Fifth Circuit's discussion of risk in Compaq, claiming that the decision turned the anti-abuse jurisprudence about risk on its head. (111) It is easy to see why commentators were incensed by the appellate courts' treatment of risk. Compaq's ability to lock in its resale price was an integral part of the transactions. Because of the large sums involved (Compaq bought nearly $900 million of Shell ADRs), even a small decline in price would have cost Compaq a great deal of money. (112) Compaq probably was not willing to take the risk of a substantial decline in the price of its Shell stock; it likely would not have engaged in the transactions without a hedge. Thus, the courts misread a key part of the strategy.
Yet the Fifth Circuit's discussion of risk has a certain logic to it. To see that logic, that discussion should not be read separately from the rest of the court's opinion. Instead, it should be read in light of the court's holding that Compaq generated a pre-tax profit from the transactions. That is because the taxpayer's risk reduction activities are not an independent basis for denying it the tax benefits that it seeks; instead, risk reduction is closely related to the issue of before-tax profit. If a transaction yields an expected before-tax loss, then the taxpayer, by eliminating all risk, ensures that it will suffer a before-tax loss. In that case, the taxpayer's risk reduction activities strengthen the government's argument that the motivation for as well as the expected (and actual) benefits from the transaction are exclusively tax benefits. Conversely, if a transaction yields a before-tax profit, then by eliminating risk the taxpayer ensures that it will enjoy a before-tax profit. In these circumstances, the employment of risk reduction techniques does not undercut the taxpayer's argument that non-tax benefits both motivate and are the actual result of the transactions. (113)
It is, then, easy to understand the court's reasoning. Having concluded that the transactions were economic arbitrages, not tax arbitrages, there is nothing suspicious about the taxpayer locking in a pre-tax profit with a collar, even one that eliminates all risk. Such a hedge in the court's view protects the taxpayer's arbitrage profits. (114) Thus, the problem with the court's reasoning is not so much that it was mistaken about risk as it was mistaken about pre-tax profit.
The calls by commentators to reinterpret the pre-tax profit test in light of the principles of anti-abuse law, to modify that test, or to abandon it altogether should go unheeded. They are unnecessary and undesirable. There is no need for such a drastic and dramatic response, or at the very least, cross-border, dividend-stripping transactions do not reveal such a need. All that is required to handle cross-border, stripping transactions properly is for the before-tax profit test to measure profit before all taxes, both explicit and implicit. Such a correction will also improve the operation of the pre-tax profit test in other areas as well.
VII. INCORPORATING IMPLICIT TAXES INTO THE PRE-TAX PROFIT TEST
For readers who accept the idea that pre-tax profit should be calculated before implicit taxes, one question remains: how difficult would it be for courts to assess economic substance by applying the pre-tax profit test before both explicit and implicit taxes? The answer to that question depends on the circumstances. In cross-border, dividend-stripping transactions involving actively traded public stock, the implicit tax is easily measured. That is because the market performs a crisp experiment that isolates that tax. (115) In addition, the measurement problem is probably not severe with tax-shelter transactions that are marketed by promoters. That is because most such tax arbitrages are usually close to zero-profit transactions before taking account of fees and expenses (and, of course, taxes). Thus, arbitrage transactions with substantial fees and expenses will generate pre-tax losses unless they also produce economic arbitrage benefits larger than the fees and expenses.
Of course, if implicit taxes are going to be incorporated into the economic substance doctrine, they cannot reasonably be limited to cross-border, dividend-stripping transactions or even to cross-border, stripping transactions. Undoubtedly there will be circumstances where it will be more difficult to measure the implicit tax. For example, the test might be difficult to administer accurately with transactions undertaken by taxpayers without incurring substantial fees and expenses. In such circumstances, the pre-tax profit will likely be close to zero, which puts pressure on the accuracy with which any implicit tax is measured. Nonetheless, it is important for courts to try. The alternative is an inaccurate pre-tax profit test that does a poor job of discriminating between harmful tax arbitrages and beneficial (or at least benign) economic arbitrages. Such a test will cause taxpayers to gravitate towards tax arbitrages that pass the test and to avoid economic arbitrages that are penalized. The latter is undesirable, but the former is the very reason for the economic substance doctrine and the pre-tax profit test. And if such transactions are not foreclosed, they will be exploited. That is what Compaq and IES Industries did.
Nonetheless, incorporating implicit taxes into the economic substance doctrine opens the door to litigants complicating and confusing every tax shelter case by drawing attention to the complexity of the U.S. tax system and to the numerous ways in which taxes affect rates of return. Thus, taxpayers will be tempted to argue that once all of these influences are accounted for their transactions will be appropriately seen as profitable before all taxes. (116) Similarly, the government will be tempted to make the converse argument (perhaps in addition to other arguments)--after accounting for all of the myriad influences of the tax system, the transactions at issue produce pre-tax losses. Therefore, in order to keep such arguments in check, I suggest that the burden of persuasion rest with the party who is arguing that implicit taxes should be incorporated into the analysis. In cases similar to Compaq and IES Industries, the government should easily be able to meet that burden. In other cases, no doubt, it will be more difficult. However, such a proposal should prevent implicit taxes from becoming a refuge for either the taxpayer or the government when no one has any idea what the impact of implicit taxes has been.
In both Compaq and IES Industries, the trial courts disallowed the tax benefit that the taxpayers sought from their cross-border, dividend-stripping transactions. On appeal, both the Fifth Circuit in Compaq and the Eighth Circuit in IES Industries reversed the trial courts' disallowance of the tax benefits. The transactions and the circuit courts' grounds for their decisions were virtually identical. In both cases, the appellate courts concluded that the cross-border, dividend-stripping transactions generated before-tax profits. The commentators were split into two camps. The smaller group--the appellate courts' defenders--believed that the transactions should be respected because the taxpayers earned pre-tax profits. The other, larger group conceded that the transactions generated pre-tax profits, but nonetheless firmly believed that the transactions were abusive tax shelters and should not have been respected. That left these commentators to argue that the pre-tax profit test, which for so long has served as the centerpiece of anti-abuse jurisprudence, should be either modified or abandoned.
This Article shows that Compaq and IES Industries do not require such a drastic change in anti-abuse jurisprudence. The circuit courts reached the wrong conclusion because the parties, the courts, and commentators all neglected implicit taxes. Implicit taxes are real taxes, just as are explicit taxes, and they should therefore be treated as taxes, which is to say eliminated before any analysis of pre-tax profit. Correctly analyzed, the cross-border, dividend-stripping transactions in Compaq and IES Industries did not produce pre-tax profits. After excluding (negative) implicit taxes, which depressed the price of foreign stock that traded cum dividends on U.S. exchanges, those transactions produced pre-tax losses.
Furthermore, the issues raised in this Article are not narrowly limited to cross-border, dividend-stripping transactions. The decisions in Compaq and IES Industries opened the door to a wide range of stripping transactions. Royalties, lease payments, and interest coupons can also be stripped to generate foreign tax credits. (117) If such issues arise in the future, the government should attack them on the grounds that the transactions generate pre-tax losses after excluding implicit taxes and not on the ground that foreign taxes are expenses, which undermines the pre-tax profit test.
Finally, in 1999, Charlotte Crane observed that tax doctrine has all but ignored implicit taxes and called for commentators, lawyers, and judges to think seriously and carefully about how tax doctrines can be improved through incorporation of implicit taxes. (118) This Article responds to that challenge by recommending in a specific and detailed manner how implicit taxes can be incorporated into the pretax profit test of anti-abuse jurisprudence. No doubt there are other areas where tax doctrine can be improved by explicit incorporation of implicit taxes.
* Theodore Warner Professor, University of Pennsylvania Law School, and Professor of Real Estate, The Wharton School, University of Pennsylvania. I thank Charlotte Crane, Mitchell Kane, William Klein, Charlene Luke, James Peaslee, Robert Peroni, Daniel Shaviro, Stephen Shay, Michael Schler, Reed Shuldiner, Kirk Stark, David Weisbach, and participants at the University of Minnesota's 2006 Tax Policy Conference--The Future of Tax Shelters--for their comments and suggestions. My acknowledgment of many of the authors whose works I cite in this Article should not be taken as a claim that they agree with me, but only of my appreciation for their willingness to discuss the issues raised by our articles. Responsibility for any remaining errors is mine. This Article and the related essay appearing in Tax Notes are dedicated to the memory of Edward Kostin, an esteemed tax professional, a valued graduate lecturer, and a gentleman. See Michael S. Knoll, Implicit Taxes and Pretax Profit in Compaq and IES Industries, 114 TAX NOTES 679 (Feb. 12, 2007)[hereinafter Knoll, Implicit Taxes]. Copyright 2007 by Michael S. Knoll. All rights reserved.
(1) The attention given to the son-of-boss transaction, which is at the center of the criminal investigations of some of KPMG's former employees, has since eclipsed that given to Compaq and IES Industries.
(2) Compaq Computer Corp. v. Commissioner, 277 F.3d 778 (5th Cir. 2001), rev'g 113 T.C. 214 (1999).
(3) IES Industries, Inc. v. Commissioner, 253 F.3d 350 (8th Cir. 2001), rev'g 1999 WL 973538 (N.D. Iowa Sept. 22,1999).
(4) E.g., Peter C. Canellos, A Tax Practitioner's Perspective on Substance, Form and Business Purpose in Structuring Business Transactions and in Tax Shelters, 54 SMU L. REV. 47, 54 (2001); David P. Hariton, The Compaq Case, Notice 98-5, and Tax Shelters: The Theory is All Wrong, 94 TAX NOTES 501 (Jan. 28, 2002) [hereinafter Hariton, Compaq Case]; David P. Hariton, Tax Benefits, Tax Administration, and Legislative Intent, 53 TAX LAW. 579, 609-13 (2000); David P. Hariton, Sorting Out the Tangle of Economic Substance, 52 TAX LAW. 235, 272-73 (1999) [hereinafter Hariton, Sorting Out]; Mitchell Kane, Compaq and IES: Putting the Tax Back into After-Tax Income, 94 TAX NOTES 1215 (Mar. 4, 2002); William A. Klein & Kirk J. Stark, Compaq v. Commissioner--Where Is the Tax Arbitrage?, 94 TAX NOTES 1335 (Mar. 11, 2002); James M. Peaslee, Creditable Foreign Taxes and the Economic Substance Profit Test, 114 TAX NOTES 443 (Jan. 29, 2007); Daniel N. Shaviro, Economic Substance, Corporate Tax Shelters, and the Compaq Case, 88 TAX NOTES 221 (July 3, 2000); Daniel N. Shaviro & David A. Weisbach, Cross-Border vs. Domestic Dividend-Stripping: An Illusory Distinction, 25 TAX NOTES INT'L 1435 (Apr. 1, 2002) [hereinafter Shaviro & Weisbach, Cross-Border]; Daniel N. Shaviro & David A Weisbach, The Fifth Circuit Gets It Wrong in Compaq v. Commissioner, 94 TAX NOTES 511 (Jan. 28, 2002) [hereinafter Shaviro & Weisbach, Fifth Circuit]; David A. Weisbach, The Failure of Disclosure as an Approach to Shelters, 54 SMU L. REV. 73, 79 (2001); Lee A. Sheppard, Should Riskless Profit Equal Economic Substance?, 94 TAX NOTES 153 (Jan. 14, 2002); Marc D. Teitelbaum, Compaq Computer and IES Industries--The Empire Strikes Back, 20 TAX NOTES INT'L 791 (Jan. 3, 2000); Alan R. Weiner, Compaq: Another View, 94 TAX NOTES 777 (Feb. 11, 2002); George K. Yin, Getting Serious About Corporate Tax Shelters: Taking a Lesson from History, 54 SMU L. REV. 209, 221-23 (2001).
(5) Klein and Stark have offered the most sophisticated and thorough non-tax justification for cross-border, dividend-stripping arbitrage. Klein & Stark, supra note 4 (arguing that Compaq and IES were engaging in economic arbitrage, not tax arbitrage).
(6) Shaviro & Weisbach, Cross-Border, supra note 4 (arguing that the arbitrage that Compaq and IES were engaging in was not economic arbitrage as Klein and Stark argue, but tax arbitrage).
(7) See, e.g., Hariton, Compaq Case, supra note 4, at 504 (conceding that the transaction generated a pre-tax profit); Kane, supra note 4, at 1217 (same); Shaviro & Weisbach, Fifth Circuit, supra note 4, at 512 (same).
(8) See, e.g., Knetsch v. United States, 364 U.S. 361 (1960).
(9) Application of the economic substance doctrine is complicated by transactions that involve either or both a substantial investment for a significant period of time and sufficient risk so that the transaction is not guaranteed to produce a certain profit or loss. See Hariton, Compaq Case, supra note 4. Neither condition applies in Compaq and IES Industries. The risk was eliminated by prearrangement and the funds were borrowed.
(10) See Shaviro & Weisbach, Fifth Circuit, supra note 4, at 512-13.
(11) The Eighth Circuit in IES Industries described the standard for whether the transaction is a sham as follows: "a transaction will be characterized as a sham if 'it is not motivated by any economic purpose outside of tax considerations' (the business purpose test), and if it 'is without economic substance because no real potential for profit exists' (the economic substance test)." IES Industries, Inc. v. Commissioner, 253 F.3d 350, 353 (8th Cir. 2001) (citation omitted). That two-part test is often called the economic substance doctrine.
(12) The sales were profitable because the purchasers had both foreign source income that would benefit from the credits as well as substantial capital gains. As such, the capital losses would not be disallowed by section 1211.
(13) E.g., Shaviro & Weisbach, Fifth Circuit, supra note 4, at 512-13. Chorvat offers a theoretical defense of such a "smell test." Terrence Chorvat, The Consistency of Inconsistency, 26 VA. TAX REV. 859 (2007).
(14) E.g., Shaviro & Weisbach, Fifth Circuit, supra note 4, at 511-12.
(15) Id. at 511.
(16) Section 901 requires a fifteen-day unhedged, holding period for stocks when the holder claims a foreign tax credit.
(17) See I.R.S. Notice 98-5, 1998-1 C.B. 334, revoked by I.R.S. Notice 2004-19, 2004-1 C.B. 606.
(18) See ILM 200620022, 2006 TNT 98-20 (May 19, 2006).
(19) The Tax Court in Compaq and the District Court in IES Industries found evidence that the sale price of the transaction was set at the same time as the purchase price. Compaq Computer Corp. v. Commissioner, 113 T.C. 214, 219 (1999); IES Industries, Inc. v. Commissioner, 1999 WL 973538, *1 (N.D. Iowa Sept. 22, 1999).
(20) The shares were in the form of American Depository Receipts (ADRs).
(21) That tax is calculated as follows: $3.4 million = $22.5 million x 15 percent, rounded to one decimal place.
(22) The United States has a worldwide tax system. For a discussion of the difference between worldwide and territorial taxation, see KIRT C. BUTLER, MULTINATIONAL FINANCE 196-97 (3d. ed. 2003).
(23) The dividend is reported as income gross of the Netherlands withholding tax.
(24) I am indebted to Charlene Luke for pointing out to me that the top corporate tax rate in 1992 was 34 percent, not 35 percent, as I implied in Knoll, Implicit Taxes, supra note *, at 681. Although this revision changes some intermediate calculations such as the one in the next footnote, most of the calculations, including all of the entries in the tables, are unchanged.
(25) Thirty four percent of $22.5 million is $7.7 million (rounded to one decimal place).
(26) Compaq's capital loss ($20.5 million) is the difference between the price it paid (including fees and expenses) for its Shell stock--$888.9 million--and the price it received when it sold its shares--$868.4 million.
(27) Compaq's $20.5 million capital loss reduced its tax liability by $7 million ($20.5 million x 34 percent).
(28) The $0.7 million increase in Compaq's tax obligation is the product of its tax rate (34%) and the additional $2 million it reported in income. That additional $2 million is the amount by which the gross dividend ($22.5 million) exceeds the sum of Compaq's capital loss ($19.1 million) plus fees and expenses ($1.4 million).
(29) I.R.C. [section] 901. The maximum amount of foreign tax that the United States will credit is 34 percent of the income received, which in this case would be $7.7 million. See I.R.C. [section] 904.
(30) Compaq Computer Corp. v. Commissioner, 277 F.3d 778, 780 (5th Cir. 2001); Compaq Computer Corp. v. Commissioner 113 T.C. 214, 218 (1999).
(31) Compaq was, in the argot of tax practice, excess limitation.
(32) As several colleagues who commented on prior drafts of this Article pointed out, Compaq also knew of the arbitrage possibility (through Twenty First Securities) and decided to pursue that opportunity. There might have been other similarly situated taxpayers who either were ignorant of the opportunity or if knowledgeable were unwilling to engage in the transactions.
(33) IES Industries, Inc. v. Commissioner, 1999 WL 973538 (N.D. Iowa Sept. 22, 1999); see also Shaviro & Weisbach, Fifth Circuit, supra note 4, at 514.
(34) Compaq Computer Corp. v. Commissioner 113 T.C. 214, 222-25 (1999).
(35) Id. at 223.
(36) Compaq's before-tax revenue of $890.9 million is reduced by its $3.4 million withholding tax payment.
(37) Compaq Computer Corp. v. Commissioner, 277 F.3d 778, 784-87 (5th Cir. 2001); IES Industries, Inc. v. Commissioner, 253 F.3d 350, 354 (8th Cir. 2001).
(38) Compaq, 277 F.3d at 784-87; IES Industries, 253 F.3d at 354.
(39) Compaq paid $887.5 million for the Shell stock, $1 million to the promoter, and $400,000 in expenses. Thus, Compaq's total expenses are $888.9 million.
(40) Compaq received $868.4 million for the sale of Shell stock and $22.5 million of dividends. Thus, Compaq's total receipts are $890.9 million.
(41) Compaq, 277 F.3d at 787.
(42) IES Industries, 253 F.3d at 356.
(43) Shaviro & Weisbach, Fifth Circuit, supra note 4, at 511-12, 516-17.
(44) Hariton, Compaq Case, supra note 4, at 501.
(45) Sheppard, supra note 4, at 153.
(46) Kane, supra note 4, at 1215.
(47) Klein & Stark, supra note 4, at 1336.
(48) Weiner, supra note 4.
(49) The government argued that the transactions in Compaq and IES Industries were pre-arranged so that is why the price drop exactly equaled the net dividend. According to the government, because of market fluctuations, the actual prices for trades among unrelated parties were slightly different.
(50) See, e.g., Shaviro & Weisbach, Fifth Circuit, supra note 4, at 515-16; Hariton, Compaq Case, supra note 4, at 503-05; Kane, supra note 4, at 1216.
(51) To simplify the discussion, assume that all bonds are issued on January 1, mature on December 31, and pay no interest in the interim. Also assume that all parties are calendar year taxpayers.
(52) That is to say, for a taxpayer in the 40 percent bracket, the return on the municipal bond, 6 percent, equals the after-tax return on the corporate bond, 6 percent (= 10% x (1-40%)).
(53) The issue price of a one-year differentially taxed bond, P, is given by the equation: P = M (1 - t) / (1 + r - t), where M is the payment at maturity, r is the aftertax interest rate of the marginal taxpayer (that is to say, the taxpayer who sets the market price of the differentially taxed bond), and t is the effective tax rate paid by the marginal taxpayer on such a bond. In this example, M is set at $1100 and r is 6 percent. For the municipal bond, t is 0. Accordingly, the issue price of the municipal bond, P, is given by the equation, P = $1100/1.06. Thus, $1037.74 [P] = $1100/1.06. At an issue price of $1037.74, a holder of a municipal bond earns 6 percent interest after tax. That is to say, $1037.74 x 1.06 = $1100.
(54) That is to say, $1000 x 1.06 = $1060.
(55) Note that the implicit tax is 6 percent higher if measured at maturity rather than issuance. Thus, the present value of the tax discounted at the 6 percent after-tax discount rate is $37.74. That is to say, $40 = $37.74 x (1.06).
(56) On a one-year bond bought for $1000, a 4 percent reduction in interest rate reduces the interest payment by $40.
(57) MYRON SCHOLES ET AL., TAXES AND BUSINESS STRATEGY: A PLANNING APPROACH 125-27 (3d ed. 2005). Note further that the 10 percent interest rate does not come from observing the municipal bond, but from the before-tax return on corporate bonds (the benchmark asset in the market in which the municipal bond trades).
(58) For the lightly taxed bond, t is 20 percent (that is to say. a taxpayer with a statutory tax rate of 40 percent would pay tax at an effective rate of 20 percent on a bond where only half of the income is subject to tax and the other half can be excluded.). Accordingly, the issue price of the lightly taxed bond, P, is given by the equation, $1023.26 [P] = $1100 (1 - .2) / (1 + .06 + .2). At such a price, the bond produces $76.75 in interest. That interest is taxed at 20 percent, which generates a tax liability of $15.35. Thus, the taxpayer is left with $1084.65 after paying taxes. That is an after-tax return of 6 percent on the purchase price of $1023.26. That is to say, $1023.26 x 1.06 = $1084.65.
(59) For a taxpayer in the forty percent bracket, the explicit tax rate on the lightly taxed bond is 15 percent, not 20 percent. The explicit tax rate is 15 percent because the tax payment is $15 (= 20% of $75 interest payment) and the total interest payment before explicit and implicit taxes is $100, not $75. See SCHOLES ET AL., supra note 57, at 125-27. Thus, for taxpayers in the 40 percent bracket, the total tax rates on lightly taxed and corporate bonds are both 40 percent, and hence they are indifferent between them. Taxpayers with marginal tax rates above 40 percent will have a higher total tax on the benchmark asset than on the lightly taxed bond and so will form the clientele for the lightly taxed bond. Conversely, taxpayers with marginal tax rates below 40 percent will have a lower total tax rate on the benchmark than on the lightly taxed bond and so will avoid the lightly taxed bond.
(60) For the heavily taxed bond, t is 80 percent (that is to say, a taxpayer with a statutory tax rate of 40 percent would pay tax at an effective rate of 80 percent on a bond where double the income is subject to tax). Accordingly, the issue price of the heavily taxed bond, P, is given by the equation, $846.15 [P] = $1100 (1 - 0.8) / (1 + .06 + 0.8). At such a price, the bond produces $253.85 in interest. That interest is taxed at 80 percent, which generates a tax liability of $203.08. Thus, the taxpayer is left with $896.92 after paying taxes. That is an after-tax return of 6 percent on the purchase price of $846.15. That is to say, $846.15 x 1.06 = $896.92.
(61) That is to say, $846.15 x 1.3 = $1100.
(62) For a taxpayer in the 40 percent tax bracket, the explicit tax rate on the bond is 240 percent, not 80 percent. The explicit tax rate is 240 percent because the tax payment (assuming a $1000 issue price) is $240 (= 80% of $300 interest payment) and the total interest payment before explicit and implicit taxes is $100, not $300. See SCHOLES ET AL., supra note 57. Thus, for such taxpayers, the total tax rates on heavily taxed and corporate bonds are both 40 percent, and hence they are indifferent between them. Taxpayers with marginal tax rates below 40 percent will have a higher total tax on the benchmark asset than on the heavily taxed bond, and so will form the clientele for the heavily taxed bond. Conversely, taxpayers with marginal tax rates above 40 percent will have a lower total tax rate on the benchmark than on the heavily taxed bond and so will avoid the heavily taxed bond.
(63) SCHOLES ET AL., supra note 57, at 2.
(64) See SCHOLES ET AL., supra note 57, at 121.
(65) Crane has forcefully argued that courts and commentators need to think long and hard about incorporating implicit taxes into legal doctrines. Charlotte Crane, Some Explicit Thinking About Implicit Taxes, 52 SMU L. Rev. 339 (1999).
(66) That is to say, the before-tax cost of Compaq's Shell shares was the $890.9 million cum dividend price excluding the $3.4 million of negative implicit tax that drove that stock's market price down to $886.5 million.
(67) Although taking account of implicit taxes reduced the pre-tax profit from the transaction, it does not change the after-tax profit. It was still $1.3 million.
(68) Recall that Compaq purchased the shares for $888.9 million (including fees and expenses, but excluding the negative implicit tax) and sold them for $868.4 million.
(69) In the extreme, had the Shell stock become worthless by the time of the sale, Compaq would have still paid the withholding tax (assuming it received the dividend).
(70) The logic behind taxing dividends in general and the dividend withholding tax in particular is, in part, that dividends generally represent income earned by the payor over a period of time. The tax, however, is imposed at a point in time for administrative convenience. That point is when the cash is available and about to leave the reach of the home country's tax authorities. The tax is not imposed because the income is earned then. The justifications for the dividend withholding tax are liquidity and administrability, rather than economic accuracy. See Klein & Stark, supra note 4, at 1339.
(71) That is to say, if there was only going to be one dividend payment, perhaps because the payor plans to switch from dividends that are subject to the withholding tax to redemptions that are not, the cum dividend price will be lower in the second case than the first by the amount of tax to be withheld on the one remaining dividend payment.
(72) Such investors include non-profit entities outside of the Netherlands, investors from countries that have territorial tax systems and thus do not credit foreign taxes, and investors from countries with worldwide tax systems, but nonetheless cannot use the foreign tax credit.
(73) Several commentators have argued that what made the dividend-stripping transactions attractive to Compaq was that the cum dividend price of the Shell stock was already reduced by the value of the Dutch withholding tax. Klein and Stark, who attribute that position to other, uncited commentators, reject it. Klein & Stark, supra note 4, at 1338 n.11. In contrast, Sheppard identifies such price suppression as the source of the tax shelter in Compaq. Sheppard, supra note 4, at 154. McDaniel et al. in their insightful discussion of Compaq go further. They argue that Compaq's Shell transactions were profitable after-tax and appeared profitable before-tax only because of such price suppression. PAUL R. MCDANIEL ET AL., FEDERAL INCOME TAXATION 1332-34 (5th ed. 2004).
(74) The nature of the tax arbitrage is identified by Klein & Stark, supra note 4, at 1336-38 (describing tax arbitrage as a result of asymmetric tax treatments across parties). Although the authors do not use the expression "clientele-based arbitrage" that is the import of their discussion.
(75) See SCHOLES ET AL., supra note 57, at 134-35. That trade-off is easy to see when an individual taxpayer borrows against her home or business and invests in municipal bonds or a university issues tax-exempt debt and purchases corporate bonds.
(76) Compaq satisfied its obligation to pay $3.4 million in Dutch withholding tax when Shell withheld that sum and remitted it to the Dutch tax authorities before writing Compaq a check for the $19.1 million net dividend.
(77) See Salina Partnership LP v. Commissioner, 80 T.C.M. (CCH) 686 (2000).
(78) An arbitrage is a transaction with no net investment that is certain to produce a profit. In a tax arbitrage, the profit comes from tax savings; in an economic arbitrage, it comes from the mispricing of at least one leg of the transaction.
(79) Scholes et al. refer to such arbitrages as "clientele-based arbitrages" as opposed to "organizational-form arbitrages." See SCHOLES ET AL., supra note 57, at 134-35.
(80) The pre-tax test can run into trouble with transactions undertaken by the taxpayer without incurring substantial fees and expenses. In such circumstances, the pre-tax profit will be close to zero, which puts pressure on the accuracy with which pre-tax profit is measured. See infra Part VII. Of course, that same problem arises with the usual formulation of the pre-tax profit test with self-enacted tax shelter transactions that do not involve implicit taxes. However, the transactions are more likely to be approved when there is implicit tax on the expense side, as in Compaq and IES Industries.
(81) Such taxpayers are said to be excess limitation because they are not using the maximum amount of foreign tax credits they are allowed by section 904.
(82) At the time, there was no provision similar to section 901(k), which requires a fifteen-day unhedged holding period to receive the foreign tax credit.
(83) Of course, Compaq would not have engaged in the cross-border, dividend-stripping transaction since it would have had an after-tax loss of about $1 million (assuming that Compaq would be allowed to the use the foreign tax credit).
(84) The classic example is the time value of money.
(85) One good example is the development of the original issue discount rules.
(86) See I.R.S. Notice 98-5, 1998-1 C.B. 334, revoked by I.R.S. Notice 2004-19, 2004-1 C.B. 606.
(87) Indeed, the definition of implicit tax in Scholes et al. mentions only the usual case of an under-taxed asset. SCHOLES ET AL. supra note 57, at 2. They mention negative implicit taxes only in a footnote. Id. at 125 n.4.
(88) Shaviro & Weisbach, Fifth Circuit, supra note 4, at 515. Their principal argument--that courts need to be flexible in attacking abusive tax shelters--is taken up later.
(89) See ILM 200620022, 2006 TNT 98-20 (May 19, 2006).
(90) See Kane, supra note 4, at 1216.
(91) The example in the text assumes that foreign taxes are considered before-tax when they are reduced as well as increased by a transaction. It is unclear how such a rule would operate in practice. In particular, Shaviro and Weisbach, because they advocate a flexible "know it when you see it" approach to tax shelters, would likely not support treating such a transaction as a tax shelter. Shaviro & Weisbach, Fifth Circuit, supra note 4, at 512-13.
(92) Id. at 515. For a more detailed and complicated example, see Peaslee, supra note 4, at 449-50.
(93) Kane, supra note 4, at 1216.
(94) See ILM 200620022, 2006 TNT 98-20 (May 19, 2006).
(95) The result does not even depend upon what effect the Dutch withholding tax has upon the stock price. If the tax had no impact so that the stock traded cum dividend at a price that reflected the gross dividend, then there would be no negative implicit tax. The before-tax purchase price would still be $890.9 million before expenses and the transaction still generates a before-tax loss of $1.4 million.
(96) Kane, supra note 4, at 1217.
(97) The $2.7 million figure ignores any tax paid by the promoter.
(98) Kane, supra note 4, at 1217. Compaq reported a profit of $2 million because it included the Dutch withholding tax in income by reporting the gross (not the net) dividend and deducted fees and expenses ($1.4 million).
(99) He provides a legal justification, by fitting the test into one of the many versions of the economic substance test, but not an explanation for why this test should be used.
(100) See Kane, supra note 4, at 1216 (criticizing Shaviro and Weisbach for suggesting that pre-tax profit should be assessed after foreign taxes because it would treat otherwise identical cross-border and domestic transactions differently).
(101) In addition, although this piece is often overlooked, the taxpayer must leave the assets in the transaction long enough to produce sufficient income to offset any expenses and generate a profit. The transactions in Compaq and IES Industries took only about an hour, so there was little opportunity for any investment (even if very large) to generate much income.
(102) Kane forcefully makes this argument. Kane, supra note 4, at 1215-16.
(103) Hariton, Sorting Out, supra note 4.
(104) Compaq Computer Corp. v. Commissioner, 113 T.C. 214, 217-18 (1999) (describing how blended purchase price set to equal sales price plus net dividend); IES Industries, Inc. v. Commissioner, 1999 WL 973538, *at 1 (N.D. Iowa Sept. 22, 1999) (describing how IES bought and resold the ADRs at prearranged prices).
(105) Hariton, Sorting Out, supra note 4; Hariton, Compaq Case, supra note 4, at 502.
(106) Shaviro & Weisbach, Fifth Circuit, supra note 4.
(108) Chorvat, supra note 13.
(109) According to Hariton, failing the pre-tax profit or changed economic position test is a necessary, but not sufficient, condition for a court to recharacterize a transaction. If the taxpayer fails the test, the transaction will still be respected and the tax benefits allowed when the tax benefits are specifically intended by the tax law. Hariton, Compaq Case, supra note 4, at 502.
(110) Compaq Computer Corp. v. Commissioner, 277 F.3d 778, 786 (5th Cir. 2001); see also IES Industries, Inc. v. Commissioner, 253 F.3d 350, 355 (8th Cir. 2001).
(111) E.g., Shaviro & Weisbach, Fifth Circuit, supra note 4, at 516.
(112) A fall of about 1.5 percent would have eliminated all of Compaq's after-tax profit.
(113) Recall that a money pump, even a small one, can be made large through leverage and replication.
(114) Compaq, 277 F.3d at 786; IES Industries, 253 F.3d at 355.
(115) If the market is liquid and deep with many trades, the negative implicit tax is the excess of the gross dividend over the stock price drop when the stock shifts from trading cum dividend to ex dividend.
(116) I am reminded of Tilly Goldstein, who won the Irish Sweepstakes and sought to defer the tax and reduce the impact of the progressive tax system on her winnings through a complex investment in bonds with different maturities. When asked why she made the investment, Tilly, a simple woman with little investment experience, responded that she was seeking to arbitrage the yield curve. The courts did not buy it and denied her the tax benefits that she sought. Goldstein v. Commissioner, 364 F.2d 734 (2d Cir. 1966).
(117) Shaviro & Weisbach, Fifth Circuit, supra note 4, at 517.
(118) Crane, supra note 65.
TABLE 1. COMPAQ'S CASH FLOWS FROM ITS TRANSACTION IN SHELL STOCK. Before After Foreign After All Proceeds from Sale Taxes Tax Taxes and Dividend: (in millions) (in millions) (in millions) Sale $868.4 $868.4 $868.4 Dividend 22.5 22.5 22.5 FTC -- -- 3.4 W/h Tax -- (3.4) (3.4) U.S. Tax -- -- (0.7) Subtotal $890.9 $887.5 $890.2 Cost of Purchase: Purchase $887.5 $887.5 $887.5 Fees 1.0 1.0 1.0 Expenses 0.4 0.4 0.4 Subtotal $889.9 $888.9 $888.9 Net Profit $2.0 ($1.4) $1.3 TABLE 2. BEFORE-TAX PRICE OF DIFFERENTIALLY TAXED BONDS.* Bond Tax Implicit Before-Tax Bond Rate Issue Price Tax Price Benchmark Single $1000.00 None $1000.00 Untaxed Untaxed $1037.74 $37.74 $1000.00 Lightly Half $1023.26 $23.26 $1000.00 Taxed Heavily Double $846.15 ($153.15) $1000.00 Taxed *Table assumes all bonds are issued so that they pay $1100 in one year at maturity; bond tax rate is expressed as a multiple of the taxpayer's tax rate; issue price is the price at which bond sells for in the market; and prices are set by taxpayers in the 40 percent bracket.
|Printer friendly Cite/link Email Feedback|
|Author:||Knoll, Michael S.|
|Publication:||Virginia Tax Review|
|Date:||Mar 22, 2007|
|Previous Article:||In defense of requiring back-flips.|
|Next Article:||Tax shelters, Dutch books, and the fundamental theorem of asset pricing.|